Geometry: making a tangent to two arbitrary circles
Moderators: gmalivuk, Moderators General, Prelates

 Posts: 2
 Joined: Sat Mar 05, 2011 12:29 am UTC
Geometry: making a tangent to two arbitrary circles
This is a problem from my high school geometry class, back in the early eighties. I wasn't able to solve it then, and it still bugs me.
Problem: Using a compass and an (unmarked) straight edge, draw a line that is tangential to two arbitrary circles.
If you simplify the problem by making the two circles have the same diameters, the problem is easy:
1) draw a line through the two centers of the circles
2) at each circle center, draw a line perpendicular to the line in step 1
3) draw a line connecting where the lines in step 2 intersect the circles
Once I make the circles have different diameters, I can't seem to figure out how to solve the problem.
One interesting thing is that it's not terribly hard to *experimentally* come up with a line that is tangential to two arbitrary circles  just keep adjusting the straight edge bit by bit until you've got it touching each circle at exactly one point. But, that wasn't the goal of the problem.
Thank you for your help!
Problem: Using a compass and an (unmarked) straight edge, draw a line that is tangential to two arbitrary circles.
If you simplify the problem by making the two circles have the same diameters, the problem is easy:
1) draw a line through the two centers of the circles
2) at each circle center, draw a line perpendicular to the line in step 1
3) draw a line connecting where the lines in step 2 intersect the circles
Once I make the circles have different diameters, I can't seem to figure out how to solve the problem.
One interesting thing is that it's not terribly hard to *experimentally* come up with a line that is tangential to two arbitrary circles  just keep adjusting the straight edge bit by bit until you've got it touching each circle at exactly one point. But, that wasn't the goal of the problem.
Thank you for your help!
Last edited by platyhiker on Tue Mar 08, 2011 1:52 am UTC, edited 1 time in total.
Re: Geometry: making a tangent to two arbitrary circles
I don't know the full solution, but have a suggestion for the first few steps.
Do your steps 13. As you said, the resulting line is not the tangent unless the circles are the same size.
Find the intersection of this line with the line that goes through the two circle centres.
The tangent line you are looking for also has to go through this intersection point. (proof using similar triangles omitted)
So now you have reduced the problem from "a tangent to 2 circles" to "a tangent to 1 circle, also going through a given point".
Do your steps 13. As you said, the resulting line is not the tangent unless the circles are the same size.
Find the intersection of this line with the line that goes through the two circle centres.
The tangent line you are looking for also has to go through this intersection point. (proof using similar triangles omitted)
So now you have reduced the problem from "a tangent to 2 circles" to "a tangent to 1 circle, also going through a given point".

 Posts: 272
 Joined: Thu Jul 09, 2009 9:26 am UTC
Re: Geometry: making a tangent to two arbitrary circles
As was pointed out to me recently when I said something similar, two arbitrary circles don't always have a common tangent. You need to make sure one circle is not fully inside the other.
Anyway, here's a way to do it http://web.mat.bham.ac.uk/C.J.Sangwin/T ... mpass.html Took a minute for the java applet to load for me.
P.S. If someone had asked me this, I'd have just lined the straightedge up with the two circles and drawn a line. Not sure why that constitutes cheating  not that different from lining the straightedge up with two points and connecting them.
Anyway, here's a way to do it http://web.mat.bham.ac.uk/C.J.Sangwin/T ... mpass.html Took a minute for the java applet to load for me.
P.S. If someone had asked me this, I'd have just lined the straightedge up with the two circles and drawn a line. Not sure why that constitutes cheating  not that different from lining the straightedge up with two points and connecting them.
Re: Geometry: making a tangent to two arbitrary circles
greengiant wrote:As was pointed out to me recently when I said something similar, two arbitrary circles don't always have a common tangent. You need to make sure one circle is not fully inside the other.
Anyway, here's a way to do it http://web.mat.bham.ac.uk/C.J.Sangwin/T ... mpass.html Took a minute for the java applet to load for me.
P.S. If someone had asked me this, I'd have just lined the straightedge up with the two circles and drawn a line. Not sure why that constitutes cheating  not that different from lining the straightedge up with two points and connecting them.
According to everyone's favourite source,
Wikipedia wrote:Each construction must be exact. "Eyeballing" it (essentially looking at the construction and guessing at its accuracy, or using some form of measurement, such as the units of measure on a ruler) and getting close does not count as a solution.
More specifically, the idea of compass and straightedge constructions is that your straightedge can only draw lines between two previously identified points, and points can only be identified by intersections of lines and/or circles. So you don't know where on the circles you're meant to line your straightedge up with to draw the tangents until you've done something to find the exact points.
pollywog wrote:I want to learn this smile, perfect it, and then go around smiling at lesbians and freaking them out.Wikihow wrote:* Smile a lot! Give a gay girl a knowing "Hey, I'm a lesbian too!" smile.

 Posts: 272
 Joined: Thu Jul 09, 2009 9:26 am UTC
Re: Geometry: making a tangent to two arbitrary circles
Yeah, that's fair enough. I get that the compass and straightedge game is played to certain official rules which don't allow this sort of thing. It still feels very restrictive though.
I'm probably being silly, but I'll try to explain. I guess I generally take the view that if you can find some way to make better use of your compass and straightedge, more power to you. I'd agree that any form of guesswork isn't on, but I don't see why lining your edge up with two circles is inherently inaccurate. I don't like the idea of saying 'you're given a compass and straightedge' then saying 'stop, you weren't meant to use it like that'  feels like reprimanding someone for a clever idea.
Maybe my problem is thinking in terms of an idealised compass and an idealised straightedge, when I should think in terms of an arcproducing machine and a machine that joins two points with a straight line.
I'm probably being silly, but I'll try to explain. I guess I generally take the view that if you can find some way to make better use of your compass and straightedge, more power to you. I'd agree that any form of guesswork isn't on, but I don't see why lining your edge up with two circles is inherently inaccurate. I don't like the idea of saying 'you're given a compass and straightedge' then saying 'stop, you weren't meant to use it like that'  feels like reprimanding someone for a clever idea.
Maybe my problem is thinking in terms of an idealised compass and an idealised straightedge, when I should think in terms of an arcproducing machine and a machine that joins two points with a straight line.
 Talith
 Proved the Goldbach Conjecture
 Posts: 848
 Joined: Sat Nov 29, 2008 1:28 am UTC
 Location: Manchester  UK
Re: Geometry: making a tangent to two arbitrary circles
You should think about the problem as using a set of axioms which you can not work out of. This is important because there are important fields of numbers and structures that are made from these kinds of axioms (e.g the constructable numbers). So although you might say "well hey, why can't I just do this?" there is normally a pretty good reason why you can't.

 Posts: 272
 Joined: Thu Jul 09, 2009 9:26 am UTC
Re: Geometry: making a tangent to two arbitrary circles
Yeah, like I said, I'm probably being silly. I understand the concept of axiomatic systems, I've actually done a fair amount of set theory/logic/model theory.
Maybe it'd be more clear to say I find it mentally jarring not to be able to use my straightedge that way. I understand it's not in the rules.
Maybe it'd be more clear to say I find it mentally jarring not to be able to use my straightedge that way. I understand it's not in the rules.

 Posts: 2
 Joined: Sat Mar 05, 2011 12:29 am UTC
Re: Geometry: making a tangent to two arbitrary circles
jaap  It turns out your suggestion is correct. Interestingly, I think did try that approach when I originally got the problem, but it didn't seem to work then  I guess I got bitten by lack of precision of my tools and drawing skills. Since your post, I tested the idea algebraically (with a specific set of circles) and found that the tangent did go through the intersection point.
Another interesting thing is that I do NOT remember enough geometry to be able to solve the problem of "draw a tangent to 1 circle, going through a given point", and I'm not at all sure I ever did.
greengiant  thank you for URL for a full solution. I don't think I could ever come up with that on my own. For anybody who wants to solution without following the link, here's my write up of how to get one of the tangents:
0) Draw your circles and mark their centers
1) Determine the difference between the two circles radii. (Call it D)
2) Draw a circle of radius D around the center of the larger circle
3) Draw a line between two circle centers
4) Determine the midpoint of the line (Call it O)
5) At O, draw a circle that goes through the center of each starting circle
6) Mark points that step 2 circle intersects with step 5 circle
7) Draw line from one of those points to center of smaller circle
8) Draw line from same point that is perpendicular to step 7 line
9) Mark where step 8 line intersects larger circle
10) At step 9 point, draw line perpendicular to step 8 line.
(The Java applet at the link covers how to get all 4 of the possible tangents.)
As for the discussion on what is allowable with straight edge and compass problems, I look at it as there is certain beauty/elegance to defining a system of rules and determining what is and is not possible within that system. For the straight edge and compass set of problems, you can do lots of neat things  duplicate angles, lengths, make parallel lines, make perpendicular lines, bisect angles and lines, and yet, there are things that are impossible  trisect an arbitrary angle, double a cube, square a circle. In college, I did a second semester of abstract algebra and near the end of semester we were able to prove the impossibility of those constructions. I don't remember any details of the proof, but I remember finding it very, very elegant and very interesting.
Thank you again for helping with a problem that has bothered me for years!
Another interesting thing is that I do NOT remember enough geometry to be able to solve the problem of "draw a tangent to 1 circle, going through a given point", and I'm not at all sure I ever did.
greengiant  thank you for URL for a full solution. I don't think I could ever come up with that on my own. For anybody who wants to solution without following the link, here's my write up of how to get one of the tangents:
0) Draw your circles and mark their centers
1) Determine the difference between the two circles radii. (Call it D)
2) Draw a circle of radius D around the center of the larger circle
3) Draw a line between two circle centers
4) Determine the midpoint of the line (Call it O)
5) At O, draw a circle that goes through the center of each starting circle
6) Mark points that step 2 circle intersects with step 5 circle
7) Draw line from one of those points to center of smaller circle
8) Draw line from same point that is perpendicular to step 7 line
9) Mark where step 8 line intersects larger circle
10) At step 9 point, draw line perpendicular to step 8 line.
(The Java applet at the link covers how to get all 4 of the possible tangents.)
As for the discussion on what is allowable with straight edge and compass problems, I look at it as there is certain beauty/elegance to defining a system of rules and determining what is and is not possible within that system. For the straight edge and compass set of problems, you can do lots of neat things  duplicate angles, lengths, make parallel lines, make perpendicular lines, bisect angles and lines, and yet, there are things that are impossible  trisect an arbitrary angle, double a cube, square a circle. In college, I did a second semester of abstract algebra and near the end of semester we were able to prove the impossibility of those constructions. I don't remember any details of the proof, but I remember finding it very, very elegant and very interesting.
Thank you again for helping with a problem that has bothered me for years!
Re: Geometry: making a tangent to two arbitrary circles
One justification for forbidding lining up the straightedge with two circles but allowing it with two points is that, with two points, you can first line up the straightedge with one, then pivot it around the point until it hits the other. That is, it's essentially two onedimensional problems.
You can do something similar with circles, first line it up to be tangent to your circle, then pivot around the center of that circle until it is tangent to the other. The difference is that you are pivoting around a point which is not on the straightedge, a less "physical" operation.
You can do something similar with circles, first line it up to be tangent to your circle, then pivot around the center of that circle until it is tangent to the other. The difference is that you are pivoting around a point which is not on the straightedge, a less "physical" operation.
Jerry Bona wrote:The Axiom of Choice is obviously true; the Well Ordering Principle is obviously false; and who can tell about Zorn's Lemma?
Re: Geometry: making a tangent to two arbitrary circles
Unless you have a compass with a right angle bend in it! (For instance, a square with a sliding pivot ring)
Any guesses what the constructible numbers are if you add such a device?
Any guesses what the constructible numbers are if you add such a device?
 jestingrabbit
 Factoids are just Datas that haven't grown up yet
 Posts: 5965
 Joined: Tue Nov 28, 2006 9:50 pm UTC
 Location: Sydney
Re: Geometry: making a tangent to two arbitrary circles
quintopia wrote:Any guesses what the constructible numbers are if you add such a device?
I'd expect them to be the same as before. We can construct a tangent to two circles, and that seems to be the only "extra" that this tool offers.
ameretrifle wrote:Magic space feudalism is therefore a viable idea.
Re: Geometry: making a tangent to two arbitrary circles
and what if you could use a trammel? still the same?
 jestingrabbit
 Factoids are just Datas that haven't grown up yet
 Posts: 5965
 Joined: Tue Nov 28, 2006 9:50 pm UTC
 Location: Sydney
Re: Geometry: making a tangent to two arbitrary circles
Off the top of my head... no idea. They only help you draw ellipses? If you require that the length of the axes are constructible, which seems fair, you only get more when you intersect ellipses, which could be reduced to the case of an ellipse and a circle by dilating a coordinate. The question is do the coordinates of an intersection of that kind represent roots of general quartics, some class of quartics, or are they something of lower degree, disguised as something trickier?
At first glance of some algebra I just scribbled out, the coordinates seem to solve quartics, but from theory its clear that they aren't general quartics (they can't be imaginary). But my gut is saying that they do allow more complicated numbers, so, plug
into alpha. Note that the nontrivial real solution has cube roots.
Answer: you get more numbers when you allow a trammel.
At first glance of some algebra I just scribbled out, the coordinates seem to solve quartics, but from theory its clear that they aren't general quartics (they can't be imaginary). But my gut is saying that they do allow more complicated numbers, so, plug
Code: Select all
http://www.wolframalpha.com/input/?i=solve%282*x^2+%2B+y^2+%3D+1%2C+%28x1%29^2+%2B+%28y1%29^2+%3D+1%29
into alpha. Note that the nontrivial real solution has cube roots.
Answer: you get more numbers when you allow a trammel.
ameretrifle wrote:Magic space feudalism is therefore a viable idea.
Who is online
Users browsing this forum: No registered users and 9 guests